<h1 id="yeah-you-get-it-right-this-clothes-is-made-of-cum.-or-it-seems-like-it...">Yeah you get it right, this clothes is made of cum. Or it seems like it… 🙃</h1><h2 id="why-i-did-it-_(-)_">Why I did it? ¯\_( ͠ᵔ ‿‿ ͡ᵔ)_/¯</h2><h3 id="it's-deer-secret">It's deer secret🦌</h3>There are 2 LoRA versions for Illustrious and 1 for Anima. Look below and find the description you need.<hr /><h1 id="version-2.1-anima">Version 2.1 Anima</h1><hr />At first, I tried using the same dataset that had been used for Illustrious. But Anima just couldn't process the old tags properly. So I had to completely overhaul the prompt structure to suit Anima's needs<h2 id="prompt">Prompt ✍</h2>This time, LoRA does not have a main tag. The main work is carried out by the following structure: <ul><li>{item of clothing} made of cum</li></ul>You must use this structure for every item of clothing (otherwise it won’t work properly). I also recommend always using the tags <code>cum covered, bukkake</code> to make the clothing more cum.You can also make the clothing transparent using the following tag:<ul><li>transparent clothes</li></ul>Sometimes it may look more like transparent latex. To fix it, use the tags <code>cum covered, bukkake</code> mentioned earlier. <h2 id="additional-helping-tags">Additional helping tags 💊</h2>There are a few tags that can help you achieve the desired result.<ul><li>cum covered, bukkake - add more cum (obviously). Can help, when clothes are not enough cum, or <code>transparent clothes</code> looks like latex. Recomends for everytime usage. You can increase the strength of these tags to increase the amount of cum (up to 2.0–3.0, then it will most likely break)</li><li>slime clothes - make clothes more liquid. Can help when clothes don't want to become cum, but sometimes can change colors a little.</li><li>white {item of clothing} made of cum - just add <code>white</code> in front of main structure. Sometimes it can help, when clothes don't want to become cum. Also it can fix color change when using <code>slime clothes</code>.</li></ul><h2 id="rare-usage-tags">Rare usage tags 💦</h2>These tags do exist, but you may not need them at all.<ul><li>puddle of cum - same as before. It controls amount of cum on floor and background. In positive prompt can add puddle of cum under girl's feet. In negative prompt it can remove puddles of cum, but sometimes also can remove cum from clothes (most likely you don't want it). Usually this tag is not needed in negative prompt, because in most cases there will be no puddle in image (or it will be small enough). </li><li>in puddle of cum - same as before. It describes situation, when part of legs is inside puddle of cum. In negative prompt this tag usually allows to avoid this situation. Using in positive prompt is a little more difficult. If you will use only this tag, most likely you will not get what you want. You need to add tags, that describe image composition (it is strange, I know). For example you can use tags "upper body", "portrait" or "close view". Puddle will be in bottom part of image, and you need to describe which part of girl will be on image. Other parts will be "in puddle of cum".</li></ul><h2 id="some-prompting-tips">Some prompting tips 🥱</h2><ul><li>If simple clothes don't work, then try use all tags at once! Use main structure, "transparent clothes" and all additional helping tags. Usually it helps. </li><li>If complex outfit don't work, then split it! Sometimes splitting outfit to few simplier pieces can help. For example: succubus cosplay -&gt; succubus lingerie + horns + tail.</li><li>If character's original clothes don't become cum, then split it! At least try. It can work, or it cannot work.</li><li>There is some combination of tags, that allows to make not clothes that looks like cum, but bukkake that looks like clothes. It’s very difficult to set up and is nothing more than a random feature (this feature hasn’t been trained by me). You need main structure for every piece of clothes, tags <code>(covered in cum, bukkake:3)</code>, <code>transparent clothes</code> and <code>holes in {clothing}</code>, for every piece of clothes. Than you need lower strenghs of min structures, until you get desired result. Difficult is that you need to find perfect strengh for main structure. Too weak, then it will not looks like clothes. Too strong, then it look more like clothes. Same thing with <code>holes in {clothing}</code> . Too much or too small will not get desired result. Also change other strenghs can help. This is a very tricky puzzle for the brightest minds. There is one example image on the model page. You can use it as a reference.</li></ul><h2 id="compatibility">Compatibility</h2><span style="background-color:rgb(26, 27, 30);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">This LoRA was trained with<span style="background-color:rgb(26, 27, 30);color:rgb(141, 141, 143);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px"> <a target="_blank" rel="ugc" href="https://civarchive.com/models/2458426/anima?modelVersionId=2945208">Anima base v1.0</a>. <span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">I tested LoRA with it and with <a target="_blank" rel="ugc" href="https://civarchive.com/models/2642932/janima?modelVersionId=2967640">JANIMA v1.0</a><span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">. Both works good. I also tried few character LoRAs and there was no serious problems.<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">I also tested it with few characters LoRAs and didn't find any issues.<hr /><h1 id="version-2.0-illustrious">Version 2.0 Illustrious</h1><hr />In second version I decided to make propt simplier. Now you need only main tag, that works on all clothes at once. Because of that LoRA works with almost all clothes. Also LoRA now have less affect on style.AND NOW IT WORKS EVEN WITH DRAGONLORD'S ARMOR (somehow)<h2 id="how-to-use-it">How to use it 🤔</h2>Use main tag and then describe clothes as usual. I also recomend to describe white color for each clothing part, like "white shirt, white short skirt". If you will describe some other color, most likely that clothing will be as usual, not cum.Also try to use additional tags.If clothes already have some associated style or color for some details, it may remains. You can try to increase some tag strenghts, but I don't guaranteer that it will completely fix this problem.<h2 id="tags">Tags ✍</h2>Main tag: cum-cl0thes<h3 id="additional-tags:">Additional tags:</h3><ul><li>transparent_clothes - makes clothes slightly transparent, looks simmilar to wet clothes. Works mostly with upper clothes, like shirts and upper dress part.</li><li>cum_covered - adds cum on naked parts of body and on clothes.</li><li>cum_puddle - helping tag, controls amount of cum on floor and background. In positive prompt can add puddle of cum under girl's feet. In negative prompt it can remove puddles of cum, but also can remove cum from clothes (most likely you don't want it). Usually this tag is not needed in negative prompt, because in most cases there will be no puddle in image.</li><li>in_puddle - describes situation, when part of legs is inside puddle of cum. In negative prompt this tag allows to avoid this situation. Using in positive prompt is a little more difficult. If you will use only this tag, most likely you will not get what you want. You need to add tags, that describe image composition (it is strange, I know). For example you can use tags "upper body", "portrait" or "close view". Puddle will be in bottom part of image, and you need to describe which part of girl will be on image. Other parts will be "in_puddle".</li></ul><h2 id="compatibility">Compatibility</h2>This LoRA was trained with WAI-illustrious-SDXL v16 but works good with all checkpoints I've tested (around 15 checkpoints).If you want to use cum-clothes with my <a target="_blank" rel="ugc" href="https://civarchive.com/models/2177159">Cum-chan LoRA</a> you don't need separated cum-clothes LoRA. Use Cum-chan v5.0 LoRA, cum-clothes are already present in it.<hr /><h1 id="version-1.0-illustrious">Version 1.0 Illustrious</h1><hr />This LoRA is old and not very good. There is already 2nd version, it's better then previous.<h3 id="tags">Tags</h3>Main tag: cum-cl0thesAdditional (but highly recomended) tag: {write-your-clothing-here} made_of_cumStrenght: 0.8-1Strenght 1 can change style to more shiny (maybe you will even like it) and add cum on objects/ Stranght 0.8 might request few attepts before good imageHelping tag: cum_pudd1e. It affects the amount of cum on floor objects and amount of dripping clothes. Place it to positive prompt to increase amounts abd plase to negative to decrease. Try different strainghts.<h3 id="trained-clothes">Trained clothes 👗</h3>LoRA was trained with (random order):<ul><li>bikini</li><li>bra</li><li>panties</li><li>dress</li><li>shirt</li><li>swimsuit</li><li>bodysuit</li><li>shorts</li><li>longsleeve</li><li>shoes</li><li>socks</li><li>hoodie</li><li>pants</li><li>vest</li><li>sweater</li><li>capelet</li><li>apron</li></ul><h2 id="but-actually">BUT ACTUALLY 🤓</h2>It can work with almost all clothings.BUT it often does not work well if a some tag already describes the material, colour, pattern, etc. Sometimes that tag can even be a characters. If a character is associated with a certain type of clothing, LoRA may not affect that clothing. For example:<ul><li>Marin Kitagawa (bikini)</li><li>Ayanami Rei (bodysuit)</li></ul><h2 id="help!-my-armor-of-dragonlord-can-not-become-cum!!!">HELP! MY ARMOR OF DRAGONLORD CAN NOT BECOME CUM!!! 😭</h2>Well, I can't help with armour. But I can give you some advices on NORMAL clothing. Try these methods (from less painful to more painful method):<ul><li>try again (help often)😉</li><li>try adding tag {color-of-clothing-that-not-become-cum}_{clothing} in negative prompt. This issue may be caused by few images in the dataset.</li><li>add tag "clothes made_of_cum" and try different strenghts of this tag</li><li>add white color to discription of clothing (sometimes can help). Example: "white skirt made_of_cum"</li><li>change strenght of LoRA or strenght of main tag (cum-cl0thes) or strenght of clothing with tag "made_of_cum". If you're lucky, it won't even affect style</li><li>change checkpoint. Seriously. Or try different strenghts of LoRA or other settings. For example, I found that for JANKU checkpoint optimal strenght is between 0.6 and 0.8.</li></ul><h2 id="compatibility">Compatibility</h2>This LoRA was trained with WAI-illustrious-SDXL and it works best with it and with v14 and v15. With other checkpoints this LoRA not always works correctly and may require additional adjustments (but there are exceptions).Relevant for Cum-chanV4.0. If you want to try to use this LoRA with my <a target="_blank" rel="ugc" href="https://civarchive.com/models/2177159">Cum-chan LoRA</a> then start with 0.5 strength on both LoRAs and then change them. LoRAs was made with simmilar (but not identical) images in datasets. Because of that both LoRAs contains same cum texture and can make cum overdrive on your image (it will be bad).For Cum-chanV5.0 you don't need this LoRA, cum-clothes are already present in this LoRA.

Cum-clothes

The dataset contains the same images as the v2.0 Illustrious, but with different prompts.The prompting rules have been adapted to suit Anima’s needs. Because of that (and some Anima magic) this LoRA create a little different results. And because of that it is v2.1.

Files with SHA256: f057dec61cfb33e989a845ee5a9dabec4b2a63ec5099cfb9541b4d2ff6ea01e2